Optimization of low-cost medium for very high gravity ethanol ...

Bioresource Technology 101 (2010) 7856–7863

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Optimization of low-cost medium for very high gravity ethanol fermentations by Saccharomyces cerevisiae using statistical experimental designs Francisco B. Pereira, Pedro M.R. Guimarães, José A. Teixeira, Lucília Domingues * IBB – Institute for Biotechnology and Bioengineering, Centre of Biological Engineering, Universidade do Minho, Campus de Gualtar, 4710-057 Braga, Portugal

a r t i c l e

i n f o

Article history: Received 12 January 2010 Received in revised form 8 April 2010 Accepted 22 April 2010 Available online 2 June 2010 Keywords: Very high gravity fermentation Corn steep liquor Ethanol Response surface methodology Saccharomyces cerevisiae

a b s t r a c t Statistical experimental designs were used to develop a medium based on corn steep liquor (CSL) and other low-cost nutrient sources for high-performance very high gravity (VHG) ethanol fermentations by Saccharomyces cerevisiae. The critical nutrients were initially selected according to a Plackett–Burman design and the optimized medium composition (44.3 g/L CSL; 2.3 g/L urea; 3.8 g/L MgSO47H2O; 0.03 g/L CuSO45H2O) for maximum ethanol production by the laboratory strain CEN.PK 113-7D was obtained by response surface methodology, based on a three-level four-factor Box-Behnken design. The optimization process resulted in significantly enhanced final ethanol titre, productivity and yeast viability in batch VHG fermentations (up to 330 g/L glucose) with CEN.PK113-7D and with industrial strain PE-2, which is used for bio-ethanol production in Brazil. Strain PE-2 was able to produce 18.6 ± 0.5% (v/v) ethanol with a corresponding productivity of 2.4 ± 0.1 g/L/h. This study provides valuable insights into cost-effective nutritional supplementation of industrial fuel ethanol VHG fermentations. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction Bio-ethanol is regarded as a promising alternative energy source, which is both renewable and environmentally friendly. During bio-ethanol production, the composition of media affects the physiological state and, consequently, the fermentation performance of the microorganism employed (Hahn-Hägerdal et al., 2005). Ethanolic fermentations with very high sugar concentrations (>300 g/L) – very high gravity (VHG) fermentations – have many advantages from an industrial point of view, resulting in reduced costs because of lower energy consumption (Thomas et al., 1996). However, these fermentations are rarely fast and complete due to physiological changes in the microbial cells. The high sugar content in the fermentation medium causes an increase in the osmotic pressure, which has a damaging effect on yeast cells. Saccharomyces cerevisiae, the yeast commonly used for ethanolic fermentations, can ferment increased amount of sugars in the medium when all required nutrients are provided in adequate amounts (Bafrncová et al., 1999). Specific nutrients, such as nitrogen, trace elements or vitamins, are required to obtain rapid fermentation and high ethanol levels, desirable to minimize capital costs and distillation energy. On a laboratory scale, media are often supplemented with peptone and yeast extract. However, such addition is not feasible in industrial fermentation processes due * Corresponding author. Tel.: +351 253 604400; fax: +351 253 678986. E-mail address: [email protected] (L. Domingues). 0960-8524/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2010.04.082

to the high costs associated. Thus, it is necessary to exploit inexpensive nutrient sources to supply all nutritional requirements for yeast growth and fermentation. Corn steep liquor (CSL), a major by-product of corn starch processing, is a low-cost source of proteins, amino acids, minerals, vitamins and trace elements and can be used as a rich and effective nutritional supplement, in particular, as replacement for yeast extract and peptone in alcoholic fermentations (Amartey and Jeffries, 1994; Kadam and Newman, 1997; Lawford and Rousseau, 1997; Seo et al., 2009; Tang et al., 2006). Moreover, the effects of metal ions on yeast cell growth and fermentation are well documented. Magnesium, calcium and zinc have been reported to influence the rate of sugar conversion and are required as cofactors for several metabolic pathways (Palukurty et al., 2008; Xue et al., 2008; Zhao et al., 2009). Furthermore, the protective effects of magnesium and calcium against ethanol stress have been extensively studied (Birch and Walker, 2000; Nabais et al., 1988). Metal ion deficiencies often occur in fermentation media (Jones and Greenfield, 1984), and studies on optimization of metal ions combinations are thus of great practical importance to improve ethanol production. Medium optimization by the classical method of changing one variable while fixing the others at a certain level is laborious and time-consuming, especially when the number of variables is large. An alternative and more efficient approach in microbial systems is the use of statistical methods. Response surface methodology (RSM) is a commonly used method to assess the optimal fermentation conditions and also an efficient statistical technique for opti-

F.B. Pereira et al. / Bioresource Technology 101 (2010) 7856–7863

mization of multiple variables with minimum number of experiments. This method has been successfully applied to optimize alcoholic fermentation (Laluce et al., 2009; Palukurty et al., 2008; Ratnam et al., 2005; Yu et al., 2009). Plackett–Burman design allows testing of the largest number of factor effects with the least number of observations, and allows random error variability estimation and testing of the statistical significance of the parameters (Plackett and Burman, 1946). The Box-Behnken is a three-level factorial design, which allows estimating and interpreting interactions between various variables at a time during the optimization process. It is suitable for exploration of quadratic responses and constructs a second-order polynomial model with very few runs (Ferreira et al., 2007). In this study, factorial design approaches were used to develop a low-cost medium based on CSL and other inexpensive nutrient sources for high-performance batch VHG ethanolic fermentations by S. cerevisiae. The supplements that significantly improved ethanol production by the laboratory strain CEN.PK 113-7D were selected according to Plackett–Burman designs and the concentrations of the key nutrient factors (CSL, urea, MgSO4 and CuSO4) were optimized using a Box-Behnken design. Furthermore, the optimized medium was compared with a reference medium containing CSL as the sole nutrient source, using the strain CEN.PK 113-7D as well as an industrial strain (PE-2). 2. Methods 2.1. Yeasts The laboratory S. cerevisiae strain CEN.PK 113-7D (van Dijken et al., 2000) was used throughout the screening and optimization experimental designs. Final comparative fermentation tests were performed also with the industrial strain PE-2 (Basso et al., 2008). Stock cultures were maintained on YPD [yeast extract 1% (w/v), bacto peptone 2% (w/v) and glucose 2% (w/v)] agar plates at 4 °C. 2.2. Media and fermentations For nutrient screening and optimization, fermentation tests were performed in basic medium (BM) consisting of 296–308 g/L glucose and 15 g/L CSL, supplemented with nutrients according to the experimental designs (Tables 2 and 4). Glucose syrup and CSL were kindly provided by a starch manufacturer (COPAM, Portugal) and autoclaved separately (121 °C, 20 min). After autoclaving, the CSL was allowed to settle for 1–2 days at 4 °C and then centrifuged (15 min at 13,100 g) to remove the insolubles. The packed sediment corresponded to 15% (w/w) of the whole CSL. The main composition of the centrifuged CSL (i.e. the supernatant after centrifugation) is shown in Table 1. Batch 1 and batch 2 were

Table 1 Composition of the centrifuged CSL (values are average ± range of duplicate analyses).

Density Humidity (%, w/w) Ash (%, w/w) Free reducing sugarsa (%, w/w) Glucoseb (%, w/w) Total Kjeldahl nitrogenc (%, w/w) Fatd (%, w/w) pH a b c d

Determined Determined Determined Determined

Batch 1

Batch 2

1.15 ± 0.00 65.4 ± 0.1 6.15 ± 0.15 4.09 ± 0.04 0.439 ± 0.014 2.25 ± 0.46 3.19 ± 0.80 3.9

1.13 ± 0.02 64.7 ± 0.0 5.97 ± 0.46 3.82 ± 0.14 0.408 ± 0.006 2.26 ± 0.04 1.95 ± 0.81 4.1

by the DNS method (Miller, 1959), using glucose as standard. by HPLC. using the Tecator Kjeltec 1026 system. using the Tecator Soxtec HT2 system.

7857

obtained from the same industrial plant approximately 3 months apart. The concentrated nutrient stock solutions were sterilized by filtration and added aseptically to the medium. A reference medium (RM) consisting of 290–330 g/L glucose and 100 g/L CSL, and 2YP medium consisting of 290–330 g/L glucose, 20 g/L yeast extract and 40 g/L peptone (i.e. the double of the yeast extract and peptone concentrations used in standard YP medium), were used for comparative studies. In all cases, the medium was aerated by stirring with a magnetic bar (length of 3 cm) at >850 rpm during 20 min before inoculating the fermentation flasks, reaching >95% of air saturation (approximately, 8 ppm of oxygen) (Munroe, 2006). Yeast for inoculation was grown in 1 L Erlenmeyer flasks filled with 400 mL of medium containing 50 g/L glucose, 20 g/L peptone and 10 g/L yeast extract. After incubation at 30 °C and 150 rpm for 24–26 h (OD600 of 7–7.5), the cell suspension was aseptically collected by centrifugation (10 min at 7500g, 4 °C) and suspended in 0.9% NaCl to a concentration of 200 mg fresh yeast/mL, to minimize the transfer of nutrients from the seed culture to the fermentation medium. The yeast cells were inoculated at about 8 mg fresh yeast/ mL into 40 mL of culture medium to start the fermentation. Fermentations were done at 30 °C and 150 rpm in 100 mL Erlenmeyer flasks fitted with perforated rubber stoppers enclosing glycerol-filled air-locks to allow exhaustion of CO2 while avoiding entrance of air. The initial pH was adjusted to 5.5 with NaOH and the final pH was higher than 3.9 in all fermentations. The progress of fermentation was followed by mass loss. Samples for analyses were taken at the beginning and end of fermentation. The fermentation experiments were stopped when the weight of the flasks did not change anymore. 2.3. Analytical procedures Yeast growth was monitored by measuring the optical density of the culture at 600 nm (OD600). Cell dry weight was determined by centrifugation (10 min at 7500g, 4 °C) of 20 mL of the yeast culture in a pre-weighed dried tube, washing of the pellet with 20 mL of distilled water, drying overnight at 105 °C and weighing. Glucose, ethanol and glycerol were analyzed by HPLC, using a Varian MetaCarb 87H column eluted at 60 °C with 0.005 M H2SO4 at a flow rate of 0.7 mL/min, and a refractive-index detector. Yeast cell number was determined with a Neubauer counting chamber and viability was determined by methylene blue staining (Mills, 1941). 2.4. Experimental design A Plackett–Burmam design was performed to screen nine independent variables selected from the literature (Azenha et al., 2000; Kadam and Newman, 1997; Nabais et al., 1988; Palukurty et al., 2008; Wang et al., 2007; Xue et al., 2008; Zhao et al., 2009) as feasible supplements in alcoholic fermentations (g/L): CSL 25, urea 1.5, (NH4)2SO4 5, MgSO47H2O 5, KH2PO4 2, ZnCl2 0.01, FeSO47H2O 0.0072, CuSO45H2O 0.075, CaCl22H2O 0.8. For each variable, the presence and absence levels of supplements were tested (all trials were performed in duplicate), resulting in 13 independent experiments (Table 2). Significant positive variable effects were considered when the reported p-values were lower than 0.05. Furthermore, a Box-Behnken design was performed using the independent variables selected as significant in the screening design. Four replicates at the centre point level were also run to check if there was a non-linear relationship between the variables and the responses, leading to a total number of 28 trials (Table 4). For predicting the optimal point, a second-order polynomial function was fitted to correlate the relationship between independent variables and response. X1, X2, X3 and X4 factors were correlated by the following equation:

7858


Table 2 Experimental and predicted values of final ethanol, CO2 produced during the initial 30 h of fermentation, final biomass and final glycerol concentrations in fermentations of BM with 308 g/L initial glucose supplemented according to the Plackett–Burman design. Runs

1–2 3–4 5–6 7–8 9–10 11–12 13–14 15–16 17–18 19–20 21–22 23–24 25–26 a b

CSL

0 0 0 0 0 0 12.5 25 25 25 25 25 25

Urea

0 0 0 1.5 1.5 1.5 0.75 0 0 0 1.5 1.5 1.5

(NH4)2SO4

0 5 5 0 0 5 2.5 0 0 5 0 5 5

MgSO4

5 0 0 0 5 5 2.5 0 5 5 0 0 5

KH2PO4

0 0 2 2 2 0 1 2 0 2 0 0 2

ZnCl2

0 0.01 0.01 0 0.01 0 0.005 0 0.01 0 0.01 0 0.01

FeSO4

0.0072 0 0.0072 0.0072 0 0 0.0036 0 0.0072 0 0 0.0072 0.0072

CuSO4

0 0.075 0 0.075 0 0.075 0.0375 0.075 0.075 0 0 0 0.075

CaCl2

0.8 0.8 0 0.8 0 0 0.4 0 0 0.8 0.8 0 0.8

Ethanol (g/L)

CO2 at 30 h (g/L)

Biomass (g/L)

Glycerol (g/L)

Exp.a

Mod.b

Exp.a

Mod.b

Exp.a

Mod.b

Exp.a

Mod.b

127.5 ± 1.1 121.5 ± 2.5 105.0 ± 1.0 127.0 ± 2.2 128.1 ± 0.6 129.8 ± 1.7 130.1 ± 0.9 130.1 ± 1.1 128.6 ± 0.3 128.0 ± 1.0 131.2 ± 0.5 130.7 ± 1.0 131.4 ± 2.4

127.4 119.1 108.2 128.7 125.7 132.2 125.9 129.2 130.3 128.9 133.6 129.1 131.4

45.4 ± 0.9 47.6 ± 0.5 67.5 ± 0.1 46.3 ± 0.1 72.6 ± 0.2 43.5 ± 0.5 70.1 ± 0.1 62.3 ± 0.0 62.2 ± 0.1 76.0 ± 0.8 78.4 ± 0.3 77.8 ± 0.3 62.1 ± 0.9

49.2 46.2 70.3 45.2 71.2 45.0 61.6 64.7 61.1 73.5 79.8 77.7 65.9

5.5 ± 0.0 4.5 ± 0.0 5.3 ± 0.3 4.8 ± 0.3 5.3 ± 0.3 4.5 ± 0.0 5.8 ± 0.3 6.5 ± 0.0 6.5 ± 0.0 6.5 ± 0.0 6.0 ± 0.0 6.5 ± 0.0 4.8 ± 0.3

5.7 4.4 5.4 4.6 5.2 4.6 5.6 6.7 6.4 6.3 6.1 6.4 4.9

8.7 ± 0.2 8.5 ± 0.0 8.8 ± 0.4 8.7 ± 0.2 10.5 ± 1.6 10.0 ± 0.0 12.0 ± 0.0 7.8 ± 0.0 9.9 ± 2.1 8.2 ± 0.0 7.4 ± 0.0 10.1 ± 2.0 9.0 ± 0.2

9.3 8.1 9.6 8.8 9.9 11.0 9.6 8.0 10.0 7.9 7.8 10.0 9.6

Exp.: experimental value (average ± range of duplicate independent biological experiments). Mod.: model-predicted value.

Table 3 Plackett–Burman design – effects of nutrient sources on the final ethanol titre. Term

Coefficient

Standard error

t-Value

p-Value

CSL Urea MgSO4 ZnCl2 (NH4)2SO4 KH2PO4 CuSO4 CaCl2 FeSO4

3.426 3.122 2.326 2.269 2.165 1.626 1.497 1.192 1.573

0.7006 0.7006 0.7006 0.7006 0.7006 0.7006 0.7006 0.7006 0.9376

4.890 4.460 3.320 3.200 3.100 2.300 2.140 1.700 1.700

0.0002 0.0004 0.0043 0.0051 0.0007 0.0338 0.0485 0.1083 0.1129

Y ¼ b0 þ b1 X 1 þ b2 X 2 þ b3 X 3 þ b4 X 4 þ b11 X 21 þ b22 X 22 þ b33 X 23 þ b44 X 24 þ b12 X 1 X 2 þ b13 X 1 X 3 þ b14 X 1 X 4 þ b23 X 2 X 3 þ b24 X 2 X 4 þ b34 X 3 X 4

ð1Þ

In Eq. (1), Y is the predicted response corresponding to the ethanol titre at the end of the fermentation process. X1, X2, X3 and X4 are independent variables, b0 is an offset term, b1, b2, b3 and b4 are linear effects and b12, b13, b14, b23, b24 and b34 are interaction terms. A screening of vitamins (selected from the yeast mineral medium described by Verduyn et al. (1992)) was done in glucose medium containing the optimal concentration of nutrients (X1, X2, X3

Table 4 Experimental values of final ethanol, CO2 produced during the initial 38 h of fermentation, final biomass and final glycerol concentrations in fermentations of BM with 296 g/L initial glucose supplemented according to the Box-Behnken design. The final ethanol concentrations predicted by the model are also shown. Run

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 a b

Factor (g/L)

Ethanol (g/L) a

CSL (X1)

Urea (X2)

MgSO47H2O (X3)

CuSO45H2O (X4)

Exp.

15 15 15 15 15 15 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 35 35 35 35 35 35

0.5 1.5 1.5 1.5 1.5 2.5 0.5 0.5 0.5 0.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 2.5 2.5 2.5 2.5 0.5 1.5 1.5 1.5 1.5 2.5

5 2.5 5 5 7.5 5 2.5 5 5 7.5 2.5 2.5 5 5 5 5 7.5 7.5 2.5 5 5 7.5 5 2.5 5 5 7.5 5

0.075 0.075 0.025 0.125 0.075 0.075 0.075 0.025 0.125 0.075 0.025 0.125 0.075 0.075 0.075 0.075 0.025 0.125 0.075 0.025 0.125 0.075 0.075 0.075 0.025 0.125 0.075 0.075

113.9 115.7 124.0 110.6 117.2 129.2 123.6 118.2 110.0 113.2 135.2 118.9 132.5 132.4 132.0 133.9 123.8 112.5 134.8 134.4 122.6 126.5 124.0 128.6 135.4 110.2 125.5 131.4

Exp.: experimental value. Mod.: model-predicted value.

CO2 at 38 h (g/L) b

a

Mod.

Exp.

111.5 121.0 120.8 112.8 117.0 128.1 122.0 122.2 109.6 114.7 133.1 116.2 132.7 132.7 132.7 132.7 124.3 112.4 133.7 134.1 120.5 128.4 122.8 130.7 134.1 113.8 122.1 131.6

71.0 71.7 88.8 71.2 77.5 86.7 81.9 88.4 77.7 80.7 95.8 86.7 93.2 93.5 93.5 92.3 93.0 85.3 96.5 96.2 89.8 94.2 89.1 95.8 98.8 92.3 96.5 99.0

Biomass (g/L) Exp. 6.0 6.5 7.0 6.0 6.5 7.5 7.5 7.0 6.5 6.5 8.5 6.5 7.5 8.5 8.0 8.0 8.0 6.5 8.0 7.5 6.5 7.5 8.5 8.5 7.0 7.0 8.0 7.5

a

Glycerol (g/L) Exp.a 8.0 7.3 8.3 9.1 8.8 9.1 8.1 7.7 8.1 7.7 8.5 8.7 7.9 8.6 8.6 8.6 7.9 8.7 8.8 8.8 9.2 8.7 8.2 8.3 8.6 7.9 8.6 8.9


and X4) predicted by the Box-Behnken model (designated optimized medium – OM) and supplemented, according to a Plackett–Burman design, with (mg/L): biotin 0.1, myo-inositol 100, pantothenic acid 5, nicotinic acid 5, thiamine 5, pyridoxine 5 and p-aminobenzoic acid 1 (total number of 26 fermentation trials). Additionally, supplementation of OM with Tween 80 (2.4 ml/L), linoleic acid (60 mg/L) and ergosterol (24 mg/L) was also evaluated. 2.5. Statistical analysis The JMP™ – The Statistical Discovery Software was used for generation and evaluation of the statistical experimental design. The optimized medium composition for ethanol production was obtained by solving the regression equation. The data from final comparative fermentations, between OM and RM were analyzed by SigmaStat 3.10 (Systat software). 2.6. Determination of fermentation parameters Ethanol conversion yield was calculated as the ratio between the maximum ethanol concentration produced and the glucose consumed (difference between the initial and residual glucose concentrations) and expressed as a percentage (%) of the theoretical conversion yield, i.e. the yield considering a production of 0.511 g of ethanol per g of glucose. Ethanol productivity was defined as the ratio between final ethanol concentration and total fermentation time (fermentation was considered to be complete when the mass loss stopped). The yeast viability was calculated as the ratio between viable (non-stained) and total cell counts. 3. Results and discussion The development of a fermentation medium based on industrial substrates is economically desirable. CSL is a nutrient rich source and has been often used as a media supplement (Amartey and Jeffries, 1994; Jorgensen, 2009; Kadam and Newman, 1997; Lawford and Rousseau, 1997; Seo et al., 2009; Tang et al., 2006). To evaluate the feasibility of using CSL as the sole nutrient source to sustain high-performance VHG fermentation, fermentations of 300 g/L glucose solutions supplemented with CSL in concentrations ranging from 5 to 150 g/L were carried out. Increasing CSL concentration in the medium up to 110 g/L led to enhanced ethanol production. The highest ethanol production (ca. 125 g/L) was observed in fermentations with 75–110 g/L CSL. A slight decrease in yield occurred when CSL concentration was raised to 150 g/L (data not shown). Nevertheless, a fermentation medium with such high CSL concentration (>75 g/L) could compromise the economical viability of industrial fermentation processes, because of the high costs associated with this supplementation. Thus, the partial replacement of CSL with other inexpensive nutrient sources was studied. As the starting condition for such medium development experiments, we have used a basic medium (BM) containing 15 g/L CSL, therefore providing a minimum level of nutrients to support fermentation. 3.1. Screening of nutrient supplements that enhance VHG fermentation Plackett–Burman design-based experiments were performed to select nutrients – metal ions, phosphate and nitrogen sources – that could replace CSL and enhance fermentation parameters. The use of additional CSL was also introduced as one of the independent variables in the screening design, since its level in BM was low. The supplements were chosen on the basis of their cost and availability with a potential industrial utilization in mind. Ta-

7859

ble 2 shows the experimental data as well the values predicted by the models constructed using four distinct response variables: final ethanol titre, CO2 produced during the initial 30 h of fermentation, final biomass and final glycerol concentrations. Regression analysis showed that the models for ethanol titre (R2 = 0.85 and adjusted R2 = 0.76), CO2 at 30 h (R2 = 0.93; adjusted R2 = 0.89) and biomass (R2 = 0.93; adjusted R2 = 0.89) are adequate, while the model for glycerol was not satisfactory. The significance of each coefficient was determined by student’s t-test. The p-value was used as indicator of the statistical significance of the test. The results for the model using the final ethanol titre as the response are presented in Table 3, showing that CSL, urea and MgSO4, whose probability values are below 0.01 (significance level higher than 99%), contributed significantly to enhancing the ethanol production in VHG fermentation. CuSO4 had also a significant positive effect on ethanol production (p < 0.05, i.e. significance level >95%) and the supplementation with CaCl2 showed a positive effect but its contribution was not significant (p-value >0.05). The others supplements (ZnCl2, (NH4)2SO4, KH2PO4 and FeSO4) added to BM medium showed a negative effect on overall ethanol production (Table 3). The CO2 produced during the initial 30 h of fermentation was used as a parameter to assess the initial fermentation rate. Besides CSL and urea, ZnCl2 and KH2PO4 showed also a significant (p < 0.01) positive effect on the initial fermentation rate, with coefficients of 3.256 and 2.655, respectively (Table S1). Conversely, CuSO4 had a strong significant (p < 0.01) negative effect (Table S1). Similar qualitative results were obtained when considering the CO2 production at 48 h (Table S2). In terms of final biomass production, only CSL had a significant positive effect, while CuSO4, urea, CaCl2, (NH4)2SO4 and ZnCl2 had significant (p < 0.01) negative impacts (Table S3). Taken together, the results of the screening experiment show that addition of more CSL had the strongest contribution to enhancing the fermentation kinetics, improving the initial fermentation rate, the final ethanol titre and the final biomass concentration. Supplementation with urea had also a positive effect on the fermentation rate and on the ethanol titre. Amino nitrogen, the principal component of CSL and urea likely explain these results. The positive effect of increased free amino nitrogen concentration has frequently been reported during optimization processes of VHG fermentation medium (see e.g. Bafrncová et al., 1999; Dragone et al., 2003; Jones and Ingledew, 1994; Ratnam et al., 2005). Besides, CSL contains many other nutrients, including trace elements and vitamins (Akhtar et al., 1997) that likely have a positive impact on fermentation. Surprisingly, supplementation with (NH4)2SO4 had a negative impact on the final ethanol titre (Table 3) and only a slight positive effect (not statistically significant) on the initial fermentation rate (Table S1). This observation indicates that the selection of the nitrogen source for VHG media affects the overall yeast fermentation performance. Addition of MgSO4 significantly increased the final ethanol titre (Table 3), which is in accordance with several reports describing a positive effect of magnesium ions on yeast ethanol tolerance and fermentation (Birch and Walker, 2000; Dombek and Ingram, 1986; Hu et al., 2003; Kadam and Newman, 1997; Wang et al., 2007). Magnesium seems to protect yeast cells during fermentation by a mechanism that results in decreased plasma membrane permeability under ethanol stress conditions (Birch and Walker, 2000; Hu et al., 2003). Supplementation with CuSO4 (0.3 mM) also had a significant positive effect on the final ethanol titre (Table 3), although it affected negatively the initial fermentation rate (Table S1) and the final biomass production (Table S3). Similarly, Azenha et al. (2000) observed that addition of copper (0.1–1 mM) to synthetic medium resulted in marked increases in ethanol production by S. cerevisiae, although the fermentations became slower.

7860


3.2. Optimization of medium for VHG fermentations A three-level four-factor Box-Behnken experimental design was performed with BM supplemented with different combinations of the variables that were selected by the Plackett–Burman design as significant to enhance the final ethanol titre. Table 4 shows the experimental data and the values predicted by the model constructed using the final ethanol titre as the response variable. By applying multiple regression analysis on the experimental data, the following second-order polynomial equation giving the ethanol titre (Y) as a function of CSL (X1), urea (X2), MgSO4 (X3) and CuSO4 (X4) concentrations was obtained:

Y ¼ 132:716 þ 3:701X 1 þ 6:331X 2 3:152X 3 7:197X 4 1:955X 1 X 2 1:156X 1 X 3 þ 0:523X 2 X 3 2:956X 1 X 4 0:924X 2 X 4 þ 1:249X 3 X 4 5:606X 21 3:607X 22 4:402X 23 6:857X 24

ð2Þ

For a good statistical model the R2 value should be close to 1.0 where a value of 0.75 indicates the aptness of the model (Niladevi et al., 2009). The regression analysis of the data showed a good aptness of the proposed model with more than 93% variability in response being explained by the proposed model (R2 value of 0.9347 and adjusted R2 value of 0.8644). This indicated that Eq. (2) was a suitable model to describe the response of the experiment to ethanol production. The analysis of variance (ANOVA) of the quadratic regression model indicated that the model was highly significant, as the F value for the model was 13.298 (p = 0.0001). The results of Box-Behnken experiments showed both positive and negative dispersion of values (Table 5). The analysis of variance (ANOVA) of experimental data showed that medium supplementation with CSL, urea, MgSO4 and CuSO4 had a strongly linear effect on the response (p < 0.01, 99% significance). Increased concentrations of CSL and urea showed a positive correlation with ethanol production. Conversely, increased MgSO4 and CuSO4 concentrations in the medium had a negative effect on the maximum ethanol titre, possibly because CSL may already contain magnesium (Kadam and Newman, 1997) and copper ions and, consequently, low concentrations may be sufficient to complement CSL addition. Furthermore, there is a relatively narrow optimum concentration for copper and other heavy metals, which at high levels have toxic effects on yeast cells (Azenha et al., 2000). The optimal concentrations of the four factors that maximize ethanol production were predicted using the optimization function (standard least squares numerical method) of the JMP™ – The Sta-

Table 5 Box-Behnken design – standardized effects of nutrient supplementation on the final ethanol titre. Factor

Coefficient

Standard error

t-Value

p-Value

Intercept CSL Urea MgSO4 CuSO4 CSL urea CSL MgSO4 Urea MgSO4 CSL CuSO4 Urea CuSO4 MgSO4 CuSO4 CSL CSL Urea urea MgSO4 MgSO4 CuSO4 CuSO4

132.72 3.7011 6.3313 3.1522 7.1970 1.9550 1.1565 0.52250 2.9557 0.92450 1.2490 5.6056 3.6072 4.4022 6.8566

1.5805 0.91250 0.91250 0.91250 0.91250 1.5805 1.5805 1.5805 1.5805 1.5805 1.5805 1.2905 1.2905 1.2905 1.2905

84.0 4.06 6.94 3.45 7.89 1.24 0.730 0.330 1.87 0.580 0.790 4.34 2.80 3.41 5.31

0.0000 0.0014 0.0000 0.0043 0.0000 0.2380 0.4773 0.7462 0.0841 0.5686 0.4436 0.0008 0.0152 0.0046 0.0001

tistical Discovery Software. CSL 44.3 g/L, urea 2.3 g/L, MgSO47H2O 3.8 g/L and CuSO45H2O 0.03 g/L were chosen as the optimal concentrations (optimized medium – OM), predicting a maximum ethanol production of 139 g/L. Eight independent fermentation runs at the above optimized conditions were carried out and an average response of 130 ± 2 g/L was achieved, which reached 94% of the predicted value by the software. The good agreement between the predicted and experimental results confirmed the validity of the model. According to the screening experiment, besides CSL and urea, ZnCl2 and KH2PO4 also improved the initial fermentation rate significantly (Table S1), which is in agreement with previous reports on the positive effects of zinc and phosphorus in alcoholic fermentations (Xue et al., 2008; Yu et al., 2009; Zhao et al., 2009). However, supplementation of OM with 0.01 g/L ZnCl2 or 2 g/L KH2PO4 did not have any effect on the fermentation kinetics (Fig. S1), indicating that the demand for zinc and phosphorus was already fully covered by other components, most likely CSL. 3.3. Screening of vitamin and lipid supplements Supplementation of OM with vitamins (biotin, myo-inositol, pantothenic acid, nicotinic acid, thiamine, pyridoxine and p-aminobenzoic acid) (Table S4) and lipids (Tween 80, linoleic acid and ergosterol) was tested, envisioning further improvements of the fermentation rate and/or ethanol production. The positive impacts of vitamins (Alfenore et al., 2002) and lipids, particularly linoleic acid (Moonjai et al., 2002), oleic acid (the main component in Tween 80) and ergosterol (Casey et al., 1984), on yeast fermentation have been reported in the literature. Analyses of variance (ANOVA) showed that supplementation of OM with vitamins did not contribute significantly (p-value >0.05) to enhance the ethanol production in VHG fermentation (Table S5). The only vitamin showing a slightly positive effect on the final ethanol titre was biotin (coefficient of 0.378) but its contribution was not statistically significant (p = 0.055) (Table S5). Similarly, no improvements of final ethanol production and fermentation kinetics were observed when OM was supplemented with unsaturated fatty acids (linoleic acid or Tween 80) or ergosterol (Fig. S2). These results indicate that the yeast’s requirement for vitamins and lipids was also fulfilled by CSL. 3.4. VHG fermentations with the optimized medium using laboratory and industrial strains The significance of this medium optimization process was studied in VHG fermentations with the laboratory strain CEN.PK 1137D and with the industrial strain PE-2, which has been isolated from bio-ethanol production facilities in Brazil and is currently one of the most widely used strains in Brazilian distilleries (Argueso et al., 2009; Basso et al., 2008). Hence, several fermentations of glucose (290–330 g/L) to ethanol were performed to compare the optimized medium (OM) with a reference medium (RM) that contained 100 g/L CSL as the sole nutrient source (Table 6). Fermentation progression is illustrated by the CO2 production (mass loss) profiles shown in Fig. 1. Under the oxygen-limiting conditions used in this study, the kinetics of ethanol production closely followed the pattern of CO2 evolution. Under VHG conditions, the final ethanol titres in RM were higher (relative increase ca. 6%) than those in standard double-strength YP medium (2YP), although the initial fermentation rates were similar in these media (Fig. 1). With strain CEN.PK 113-7D, the maximum ethanol concentration in fermentations with approximately 290–300 g/L initial glucose was significantly higher (n = 8, p-value